1. Field of the Invention
The invention relates to a method and apparatus for rapid in-process, automated analysis resulting in characterization by direct measurement of relevant chemical constituents of a process solution. More specifically, the method preferably uses a combination of simultaneous measurements utilizing the techniques of In-Process, Isotope Dilution Threshold and Quantification Measurement Mass Spectrometry (IPMS) and Speciated Isotope Dilution Mass Spectrometry (SIDMS) to identify and quantify chemical reagents and their reaction products in a process solution. In addition, this method also uses a technique of quantitation by internal standard. The apparatus preferably consists of a sample handling instrument for Automated Analysis in Fluid based Processing (AAFP), coupled to an In-Process, Atmospheric Pressure Ionizer, Mass Spectrometer (IP-API-MS) that is configured to identify and quantify relevant chemical constituents of a liquid sample.
2. Cross Reference to Related Documents
The present application is related to (a) issued U.S. Pat. No. 5,414,259, issued to Howard M. Kingston on May 9, 1995, incorporated herein in its entirety by reference, and to (b) co-pending patent application Ser. No. 09/015,469, filed Jan. 29, 1998, also incorporated in its entirety by reference, and to (c) a co-pending patent application filed Dec. 4, 2001, bearing Ser. No. 10/004,627, which is also incorporated herein in its entirety by reference, and to (d) U.S. Ser. No. 10/086,025 filed Feb. 28, 2002, which is a non-provisional application based on co-pending provisional patent application Ser. No. 60/305,437, filed Jul. 13, 2001, also incorporated in its entirety by reference.
3. Description of the Prior Art
Characterization and maintenance of many process solutions is critical to the functioning of that solution. Information provided during characterization can enable the process engineer to assess the quality of the process and adjust the constituents to optimize the results. Frequently, knowledge of all relevant species is critical to the understanding and control of the chemical mechanisms being employed. The ability to determine all relevant species simultaneously and in real-time has been an unsatisfied goal for chemical process management. One such example is that of monitoring the concentration of additives and reagents in a metal electroplating (electrodeposition of electroless deposition) bath. The properties of a deposited metal are affected significantly by the inclusion of organic additives in the deposition solution. Historically these additives have been assigned functions such as ‘accelerators’, ‘suppressors’, ‘levelers’ and ‘brighteners’—labels that describe their effect on the rate of the electrodeposition of the metal and its morphology. In addition the properties of the metal plate will be affected by the concentration of the metal being plated. During plating the metal in solution will be reduced unless replenished such as by use of a soluble anode or by addition of metal salts to the solution. Additional reagents perform the function of electrolytes to maintain the conductivity of the solution, acids or bases are added to balance the solution pH, and other constituents are added to enhance the effect of the organic additives. The balance of these ingredients is even more critical in the semiconductor industry where metal deposition into high-aspect ratio features must be performed without defects. As the architecture of the integrated circuit continues to shrink effective metal interconnect depositions require stringent plating conditions.
The most basic approach in maintaining proper plating conditions is to replenish those plating components which are known to be depleted during deposition according to the rate of deposition. While this method can result in a uniform rate of metal deposition, the resulting film has been shown in one study to contain increasing contamination from carbon and sulfur atoms produced during the breakdown of the organic additives. Reduction of film contaminants is frequently achieved by either of two processes, (i) “batch dumping”, or (ii) “bleed and feed” in which a percentage of the plating solution is removed and replenished with fresh solution. The percentage of the solution that is removed in a “bleed and feed” process and the frequency of “batch dumping” are generally derived by experience. Both methods of solution control are wasteful and result in deposited films with a steady state level of contamination. The level of contamination in the film is still subject to change with significant fluctuations in process contamination.
In the semiconductor industry contamination is controlled by higher bleed-and-feed rates with a “remove-and-reconstitute” approach having recently been suggested. The remove-and-reconstitute step involves complete removal of all organic compounds followed by addition of fresh additives prior to recycling the base plating solution. Unfortunately, this approach provides no informative data regarding the plating process, and as unexpected results seem the norm when ECP [electrochemical plating] chemistries are tweaked—clearly, precise, real time, direct measurement of ECP chemistries are highly desired.
A number of other prior arts are used which can provide partial information about the constituents of an electroplating bath, but are susceptible to interferences, are incomplete, and are slow.
The most common approach to monitoring an electroplating or electroless plating bath is by electrochemical techniques. Electrochemical techniques use indirect methods of monitoring the concentrations of the organic additives. Direct measurement of the additive or byproduct is not possible unless they are electrochemically active within the potential range of the detector and no interferences are present. One of the earliest electrochemical techniques, referred to as Cyclic Voltammetric Stripping (CVS), involves placing an inert working, or indicator, electrode in the plating solution and applying a potential sufficient to deposit the metal onto the electrode surface. The potential is then shifted more positive to the point where the plated metal is oxidized, or stripped, off of the electrode surface. The corresponding current of the stripping wave is integrated to get the total charge, and the charge determines the quantity of the metal that was deposited. See U.S. Pat. No. 4,132,605. This deposition quantity relates indirectly to the efficiency of the plating solution, but provides very little information about the constituents of a complex plating mixture. In addition, the working electrode surface tends to become contaminated by additive byproducts and eventually sensitivity is lost with continuous use. Follow up improvements to this prior art simply extended the working life of the indicator electrode for a short time by adding a cleaning step to the process by introducing pulsed CVS techniques.
An improved technique was suggested in U.S. Pat. No. 4,917,777 in which an open circuit pause step was inserted after the cleansing step in which contaminants were removed by a mechanism not fully understood. See U.S. Pat. No. 4,917,777. These pulsed CVS improvements extended the active life of the indicator electrodes; however, exact information regarding the characterization of additives and degradation byproducts remained unknown.
A number of other prior art disclosures attempt to relate the concentration of additives to the kinetic behavior of the deposition process. In U.S. Pat. No. 4,324,621 an attempt to characterize organic additives was made by measuring the overpotential between reference and indicator electrodes at the point at which deposition begins. The overpotential required for deposition is affected by the organic additives, electrolytes, and impurities in the plating solution. By measuring this potential difference between the overpotential and the reversible metal reduction potential with varying concentrations of organic additives and correlating these with the physical condition of the film an indirect method of characterizing the plating solution was said to be established. A version of this technique is presented in U.S. Pat. No. 4,479,852 where the indicator electrode is first plated with the deposition metal and the equilibrium potential serves as the reference for the overpotential measurement. These techniques provided indirect data about the condition of a plating solution and are best used in plating solutions with no more than one additive. In solutions with multiple additives and impurities—each contributing in a different way to the plating efficiency—the foregoing techniques are incapable of providing the depth of information needed for proper bath maintenance. Evidence of this difficulty is provided in U.S. Pat. No. 5,192,403 where it is pointed out that when the ratio of two additives change the standard CVS technique is no longer applicable. The solution presented in this patent is to measure one of the two additives by a different technique (HPLC) and apply this result to the results obtained electrochemically.
In order to provide more information regarding the condition of a process bath with multiple organic additives and impurities, complex and slow calibration schemes are generally necessary. In U.S. Pat. No. 5,192,403 the usual calibration scheme is described in which a standard is added to a stock solution multiple times to generate a traditional calibration curve and then several aliquots of the unknown plating bath are added to an additional stock solution to obtain a similar curve for the unknown. This process is time consuming and can take 2 hours to provide results. Even moderately slow changes in additive or impurity will not be detected. To respond to this problem recent improvements to prior art electrochemical techniques involve accelerating the measurement speed. For example, in U.S. Pat. No. 5,182,131 the improvement suggested is a combination of determining the concentration of the consumable ingredient by a calculated method and is coupled with intermittent measuring of the consumable compound to verify the calculated values. This technique requires the surface area of the deposition surface to be known. Reliance on calculated values will work only if the rates of consumption are well established. This technique does not serve to monitor or identify by-products, or to quantify those additives that are not consumed at predictable rates.
Another method attempting to increase the reporting speed is presented in U.S. Pat. No. 6,280,602. Here the electrochemical reporting time is reduced by keeping the reference electrode exposed to the plating bath solution to be tested. This is done to shorten the equilibration time required prior to the measurement. However, each of the standard steps of preparing a basis solution, preparing a calibration solution, providing a calibration curve for each component to be measured, and comparing the slopes of both calibration curve and unknown sample curve need to be performed. While an improvement to existing techniques, the complex sample preparation and calibration still remains the limiting step in the rapidity of the measurement. An additional goal of Robertson's U.S. Pat. No. 6,280,602 was to provide an increase in the precision of existing methods to a level of preferably less than 10 percent.
Recent publications have introduced the use of chromatography to separate the individual additives before detection in order to make quantitative measurements of the additives and impurities. Analysis of Copper Plating Baths Suppressors and Levelers, B. Newton, et al., Proc. Electrochem Soc., V2000-27, page 1, December, 2000. Chromatography serves to separate the additives and impurities from other species that may interfere with their detection. Problems with interferences are not eliminated, only minimized using this technology. In Newton's work it is noted that the excess of sulfuric acid and copper sulfate in solution obscures the detection of some of the additives and byproducts. The use of chromatographic methods is time-consuming and produces delays in data reporting similar to the electrochemical methods mentioned previously. The solution to be tested must be spiked with standard organic additives of known concentration both to identify each additive and to generate a calibration curve in order to quantify them. Each separation step requires an elution time of ca. 20 minutes for completion. In addition, the chromatographic technique does not provide data that make identification of unexpected by-products or contaminants possible. Identification of unknown compounds whose presence is not predicted by an understanding of the process necessitates a different method to be used to identify the compounds followed by synthesis and spiking the sample. Assessment of the decomposition of the high molecular weight polymers used in an acid plating bath cannot be achieved using the robust chromatographic materials required in Newton.
The use of isotopes as part of an innovative process necessitates the examination of other applications that monitor similar components or that use novel methods to achieve these measurements. Isotopes have been used to measure components of interest. U.S. Pat. No. 4,975,378 discloses a specific method of indirect detection in chromatography. This method uses radioactive elements made into derivatives of the analyte of interest to be detected. The non-labeled analyte of interest and the labeled radioactive analyte of interest exit the chromatographic column together and the radioactive signature of the analyte of interest signals the arrival of the chromatographic fraction containing the analyte of interest. This is an indirect detection method. This disclosure is very specific and limited to column chromatography and indirect detection.
U.S. Pat. No. 5,696,378 demonstrates the determination of chlorine based on flame infrared emission spectrometry (ID-FIRE) using isotope dilution with detection by the infrared technique. The use of infrared was seen as an advantage as compared with sample preparation used in mass spectrometry. The mixture of chlorine and the isotopically enriched chlorine spike are mixed and excited thermally to permit infrared measurement of the isotopes. Thermal excitation could alter or destroy many of the species that are being analyzed in complex material. While this method works for isolated chlorine, it is not suited for chlorine in complex material combinations or for identification in fragile structures. This is fundamentally very different from mass spectrometric measurements where sample preparation is not a thermal process and where ionization uses very different mechanisms that are integral to the process of measurement.
In summary, prior art methods and apparatus for measuring and quantifying the chemical constituents consist mostly of indirect methods that are slow, inaccurate, and provide only some of the information needed for optimization of process solutions. Most modern process monitoring tools use various combinations of the aforementioned prior art in order to provide as much information as can be gathered. No one prior art method will do all additives in a short analysis time with good sensitivity.